Traffic Information Real-Time Monitoring Based on A Short-Long Term Algorithm

In this paper, we propose an approach for forecasting traffic flow, in which a short-term forecasting scheme is combined with a long-term forecasting scheme for enhancing power of the forecasting model in the recent trend tracing and the "average" pattern identifying. We employ a double exponential smoothing (DES) model and a support vector machines (SVM) model for a short-term scheme, and a Markov forecasting model for a long-term scheme. Additionally, the combined forecasting model is applied to forecasting the traffic flow of the Route 23 in Nagoya-shi, Aichi-ken, Japan.